be2net: support multiple TX queues

This patch provides support for multiple TX queues.
Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@conan.davemloft.net>

drivers/net/benet/be.h

@@ -87,6 +87,7 @@ static inline char *nic_name(struct pci_dev *pdev)
 
 #define MAX_RSS_QS		4	/* BE limit is 4 queues/port */
 #define MAX_RX_QS		(MAX_RSS_QS + 1) /* RSS qs + 1 def Rx */
+#define MAX_TX_QS		8
 #define BE_MAX_MSIX_VECTORS	(MAX_RX_QS + 1)/* RX + TX */
 #define BE_NAPI_WEIGHT		64
 #define MAX_RX_POST 		BE_NAPI_WEIGHT /* Frags posted at a time */
@@ -170,7 +171,6 @@ struct be_tx_stats {
 	u32 be_tx_reqs;		/* number of TX requests initiated */
 	u32 be_tx_stops;	/* number of times TX Q was stopped */
 	u32 be_tx_wrbs;		/* number of tx WRBs used */
-	u32 be_tx_events;	/* number of tx completion events  */
 	u32 be_tx_compl;	/* number of tx completion entries processed */
 	ulong be_tx_jiffies;
 	u64 be_tx_bytes;
@@ -184,6 +184,7 @@ struct be_tx_obj {
 	struct be_queue_info cq;
 	/* Remember the skbs that were transmitted */
 	struct sk_buff *sent_skb_list[TX_Q_LEN];
+	struct be_tx_stats stats;
 };
 
 /* Struct to remember the pages posted for rx frags */
@@ -319,8 +320,8 @@ struct be_adapter {
 
 	/* TX Rings */
 	struct be_eq_obj tx_eq;
-	struct be_tx_obj tx_obj;
-	struct be_tx_stats tx_stats;
+	struct be_tx_obj tx_obj[MAX_TX_QS];
+	u8 num_tx_qs;
 
 	u32 cache_line_break[8];
 
@@ -391,7 +392,7 @@ struct be_adapter {
 extern const struct ethtool_ops be_ethtool_ops;
 
 #define msix_enabled(adapter)		(adapter->num_msix_vec > 0)
-#define tx_stats(adapter)		(&adapter->tx_stats)
+#define tx_stats(txo)			(&txo->stats)
 #define rx_stats(rxo)			(&rxo->stats)
 
 #define BE_SET_NETDEV_OPS(netdev, ops)	(netdev->netdev_ops = ops)
@@ -405,6 +406,10 @@ extern const struct ethtool_ops be_ethtool_ops;
 	for (i = 0, rxo = &adapter->rx_obj[i+1]; i < (adapter->num_rx_qs - 1);\
 		i++, rxo++)
 
+#define for_all_tx_queues(adapter, txo, i)				\
+	for (i = 0, txo = &adapter->tx_obj[i]; i < adapter->num_tx_qs;	\
+		i++, txo++)
+
 #define PAGE_SHIFT_4K		12
 #define PAGE_SIZE_4K		(1 << PAGE_SHIFT_4K)
 

drivers/net/benet/be_ethtool.c

@@ -52,12 +52,7 @@ static const struct be_ethtool_stat et_stats[] = {
 	{NETSTAT_INFO(tx_errors)},
 	{NETSTAT_INFO(rx_dropped)},
 	{NETSTAT_INFO(tx_dropped)},
-	{DRVSTAT_TX_INFO(be_tx_rate)},
-	{DRVSTAT_TX_INFO(be_tx_reqs)},
-	{DRVSTAT_TX_INFO(be_tx_wrbs)},
-	{DRVSTAT_TX_INFO(be_tx_stops)},
-	{DRVSTAT_TX_INFO(be_tx_events)},
-	{DRVSTAT_TX_INFO(be_tx_compl)},
+	{DRVSTAT_INFO(be_tx_events)},
 	{DRVSTAT_INFO(rx_crc_errors)},
 	{DRVSTAT_INFO(rx_alignment_symbol_errors)},
 	{DRVSTAT_INFO(rx_pause_frames)},
@@ -111,6 +106,16 @@ static const struct be_ethtool_stat et_rx_stats[] = {
 };
 #define ETHTOOL_RXSTATS_NUM (ARRAY_SIZE(et_rx_stats))
 
+/* Stats related to multi TX queues */
+static const struct be_ethtool_stat et_tx_stats[] = {
+	{DRVSTAT_TX_INFO(be_tx_rate)},
+	{DRVSTAT_TX_INFO(be_tx_reqs)},
+	{DRVSTAT_TX_INFO(be_tx_wrbs)},
+	{DRVSTAT_TX_INFO(be_tx_stops)},
+	{DRVSTAT_TX_INFO(be_tx_compl)}
+};
+#define ETHTOOL_TXSTATS_NUM (ARRAY_SIZE(et_tx_stats))
+
 static const char et_self_tests[][ETH_GSTRING_LEN] = {
 	"MAC Loopback test",
 	"PHY Loopback test",
@@ -253,17 +258,15 @@ be_get_ethtool_stats(struct net_device *netdev,
 {
 	struct be_adapter *adapter = netdev_priv(netdev);
 	struct be_rx_obj *rxo;
+	struct be_tx_obj *txo;
 	void *p = NULL;
-	int i, j;
+	int i, j, base;
 
 	for (i = 0; i < ETHTOOL_STATS_NUM; i++) {
 		switch (et_stats[i].type) {
 		case NETSTAT:
 			p = &netdev->stats;
 			break;
-		case DRVSTAT_TX:
-			p = &adapter->tx_stats;
-			break;
 		case DRVSTAT:
 			p = &adapter->drv_stats;
 			break;
@@ -274,6 +277,7 @@ be_get_ethtool_stats(struct net_device *netdev,
 				*(u64 *)p: *(u32 *)p;
 	}
 
+	base = ETHTOOL_STATS_NUM;
 	for_all_rx_queues(adapter, rxo, j) {
 		for (i = 0; i < ETHTOOL_RXSTATS_NUM; i++) {
 			switch (et_rx_stats[i].type) {
@@ -285,11 +289,21 @@ be_get_ethtool_stats(struct net_device *netdev,
 								rxo->q.id;
 				break;
 			}
-			data[ETHTOOL_STATS_NUM + j * ETHTOOL_RXSTATS_NUM + i] =
+			data[base + j * ETHTOOL_RXSTATS_NUM + i] =
 				(et_rx_stats[i].size == sizeof(u64)) ?
 					*(u64 *)p: *(u32 *)p;
 		}
 	}
+
+	base = ETHTOOL_STATS_NUM + adapter->num_rx_qs * ETHTOOL_RXSTATS_NUM;
+	for_all_tx_queues(adapter, txo, j) {
+		for (i = 0; i < ETHTOOL_TXSTATS_NUM; i++) {
+			p = (u8 *)&txo->stats + et_tx_stats[i].offset;
+			data[base + j * ETHTOOL_TXSTATS_NUM + i] =
+				(et_tx_stats[i].size == sizeof(u64)) ?
+					*(u64 *)p: *(u32 *)p;
+		}
+	}
 }
 
 static void
@@ -312,6 +326,13 @@ be_get_stat_strings(struct net_device *netdev, uint32_t stringset,
 				data += ETH_GSTRING_LEN;
 			}
 		}
+		for (i = 0; i < adapter->num_tx_qs; i++) {
+			for (j = 0; j < ETHTOOL_TXSTATS_NUM; j++) {
+				sprintf(data, "txq%d: %s", i,
+					et_tx_stats[j].desc);
+				data += ETH_GSTRING_LEN;
+			}
+		}
 		break;
 	case ETH_SS_TEST:
 		for (i = 0; i < ETHTOOL_TESTS_NUM; i++) {
@@ -331,7 +352,8 @@ static int be_get_sset_count(struct net_device *netdev, int stringset)
 		return ETHTOOL_TESTS_NUM;
 	case ETH_SS_STATS:
 		return ETHTOOL_STATS_NUM +
-			adapter->num_rx_qs * ETHTOOL_RXSTATS_NUM;
+			adapter->num_rx_qs * ETHTOOL_RXSTATS_NUM +
+			adapter->num_tx_qs * ETHTOOL_TXSTATS_NUM;
 	default:
 		return -EINVAL;
 	}
@@ -457,10 +479,10 @@ be_get_ringparam(struct net_device *netdev, struct ethtool_ringparam *ring)
 	struct be_adapter *adapter = netdev_priv(netdev);
 
 	ring->rx_max_pending = adapter->rx_obj[0].q.len;
-	ring->tx_max_pending = adapter->tx_obj.q.len;
+	ring->tx_max_pending = adapter->tx_obj[0].q.len;
 
 	ring->rx_pending = atomic_read(&adapter->rx_obj[0].q.used);
-	ring->tx_pending = atomic_read(&adapter->tx_obj.q.used);
+	ring->tx_pending = atomic_read(&adapter->tx_obj[0].q.used);
 }
 
 static void